Pathfinding and Error Correction by Retinal Axons Lara D. Hutson, Chi-Bin Chien  Neuron  Volume 33, Issue 2, Pages 205-217 (January 2002) DOI: 10.1016/S0896-6273(01)00579-7

Figure 1 Growth Cone Morphology Differs between Wild-Type and astray (A) Lateral view of 36 hpf embryo (top) shows the axes used throughout this paper: R, rostral; C, caudal; D, dorsal; V, ventral. Schematic diagram (bottom) shows the retinal axon pathway, with numbers 1–4 indicating locations of growth cones shown in (B) and (C). DR, dorsorostral; VC, ventrocaudal. (B and C) Representative wild-type (wt) (B) and ast (C) growth cones. (D–F) Growth cone morphology as a function of location (midline = 0 μm) and overall (“all”). (D) Total filopodial length of wild-type growth cones (black bars) varies significantly with location (ANOVA, p = 0.0025) and is greatest between −5 and 15 μm. ast growth cones (gray bars) are significantly more complex than wild-type at several locations and overall, but significantly less complex than wild-type at the midline. (E) Number of wild-type filopodia varies significantly with location (ANOVA, p = 0.002), and is highest at the midline. ast growth cones have significantly more filopodia than wild-type at several locations and overall. (F) wild-type growth cone area does not vary significantly with location. ast growth cones are larger than wild-type at several locations and overall. n for each interval, left to right: wt, 10, 4, 11, 6, 13, 8, 9, 12, 5; ast, 5, 13, 11, 5, 8, 11, 6, 10, 4. (G) Growth rates determined from time-lapse imaging. Wild-type growth rate varies significantly with location (ANOVA, p = 0.03) and is slowest between −5 and +25 μm. ast growth rates do not differ significantly from wild-type. n for each interval: wt, 3, 5, 5, 6, 6, 8, 7, 5, 5; ast, 6, 6, 7, 7, 6, 5, 5, 4, 4. Data shown as mean ± SEM. *p < 0.05, **p < 0.005, ***p < 0.0001 (wt versus ast, two-tailed Student's t test). Neuron 2002 33, 205-217DOI: (10.1016/S0896-6273(01)00579-7)

Figure 2 Axon Misrouting in Wild-Type and astray Rostral (A–D) and ventrorostral (E and F) views of RGC axon projections; DiI-labeled eyes at left. DIC images show that the midline is centered in (A)–(E) and slightly right of center in (F). (A) Normal wild-type, 42 hpf. Axons grow in a fascicle across the midline and project dorsally toward the contralateral tectum. (B) Normal wild-type, 48 hpf. The number of axons has increased. (C) Errors in wild-type, 48 hpf. One axon misroutes at the midline, projecting ipsilaterally (arrowhead). (D) ast projection, 42 hpf. Most axons are on the normal pathway, but one axon misroutes soon after exiting the eye (closed arrowhead) and another misroutes in the contralateral optic tract (open arrowhead). (E) ast, 42 hpf. Several axons follow the normal pathway until the midline. A single axon misroutes at the midline, projecting into the contralateral eye (arrowhead). (F) ast, 36 hpf. Two axons exit the eye together and follow the normal pathway for 25 μm (closed arrowhead). One then continues normally to the midline, while the other misroutes dorsorostrally. A third axon misroutes immediately after exiting the eye (open arrowhead) and crosses the midline caudal to the normal optic chiasm. (A–F) Scale bar, 25 μm. (G) Locations of all errors in embryos from 36 to 72 hpf. Wild-type errors (black bars) occur primarily at the midline and also very rarely from −45 to −35 μm and from +35 to +45 μm. ast errors (gray bars) occur most often at the midline, but also at a low rate throughout the pathway. (H) Frequency of all errors as a function of age. In wild-type, errors are eliminated over time, while in ast, errors persist or increase. (I) Frequency of ipsilateral projection errors as a function of age. Ipsilateral errors are eliminated in both wild-type and ast. Data in (H and I) is shown as mean ± SEM of three clutches, each with at least 18 ast or 31 wild-type embryos. *p < 0.05. Neuron 2002 33, 205-217DOI: (10.1016/S0896-6273(01)00579-7)

Figure 3 Time-Lapse Imaging of a Normal Wild-Type Growth Cone Crossing the Ventral Forebrain (Left inset) Lateral view of embryo with arrow indicating the ventrorostral view used for all time-lapse imaging. (Right inset) Schematic of pathway as viewed in time-lapse, with box indicating imaged area. Rostrodorsal is up; black arrowheads below each column indicate the midline. A typical growth cone navigates across the ventral forebrain and turns caudally at the midline (60′), extending many filopodia (arrowhead, 90′). Soon after crossing the midline, it turns dorsally (135′), becoming quite simple as it extends toward the contralateral diencephalon. (See also Supplemental Movie S1 online at http://www.neuron.org/cgi/content/full/33/2/205/DC1.) Neuron 2002 33, 205-217DOI: (10.1016/S0896-6273(01)00579-7)

Figure 4 Wild-Type Axons Make Pathfinding Errors (A) Small-scale error. A single misrouted growth cone (arrowhead at 0′) deviates from a growing fascicle just after crossing the midline. It begins to show evidence of collapse at 36′ (arrowhead) and eventually retracts completely. (B) Ipsilateral projection error. A single growth cone navigates to the midline, lingering there and extending filopodia for over an hour (75′–195′). Two prominent filopodia are evident between 180′ and 195′, one directed contralaterally and one ipsilaterally. The contralaterally directed filopodium eventually broadens and collapses (closed arrowhead, 195′). The growth cone fills the ipsilaterally directed filopodium and projects into the ipsilateral optic tract (closed arrowhead, 225′). A second growth cone (open arrowhead, 90′) extends along the first axon until the midline (165′), whence it projects normally into the contralateral optic tract (open arrowheads, 210′ and 240′). (See also Supplemental Movies S2 and S3 online at http://www.neuron.org/cgi/content/full/33/2/205/DC1.) Neuron 2002 33, 205-217DOI: (10.1016/S0896-6273(01)00579-7)

Figure 5 astray Errors Are More Frequent and Rarely Corrected (A) Two closely apposed ast growth cones project toward the midline, reaching it at 60′. Here, the upper growth cone leaves the normal pathway, projecting dorsally and rostrally (out of the page). Simultaneously, the lower growth cone misroutes caudally (into the page). Dashed line indicates the approximate normal pathway. (See also Supplemental Movie S4 online at http://www.neuron.org/cgi/content/full/33/2/205/DC1.) (B) Comparison of wild-type (wt) and ast, showing frequencies of corrected errors (black bars) and uncorrected errors (gray bar). 17% of all wild-type growth cones made pathfinding errors during time-lapse, while 53% of all ast growth cones made pathfinding errors (χ2, p = 0.0075). All wild-type errors were eventually corrected,while only 11% of ast errors were corrected (Fisher exact test, p = 0.0009). Neuron 2002 33, 205-217DOI: (10.1016/S0896-6273(01)00579-7)

Figure 6 astray Function Is Required without Midline Crossing Dorsal views of retinotectal projections labeled at 5 dpf with DiI (red) and DiO (green). Rostral is up. Arrowheads indicate the midline of the ventral diencephalon, where the optic chiasm should form. (A) Schematic diagram of wild-type larva. (B) Wild-type projection. All axons project across the chiasm to the contralateral tecta. (C) ast projection. All axons cross the ventral midline, and some recross the dorsal midline in the posterior commissure (PC), resulting in innervation of ipsilateral as well as contralateral tecta. Some axons project caudally into the ventral hindbrain (VHB), and others project across the midline rostral to the optic chiasm (arrow). (D) bel projection. Axons do not cross the midline where the chiasm would normally form, and project instead to the ipsilateral optic tecta. (E) Presumptive ast/bel projection. Few, if any axons cross the midline where the chiasm should form (arrowhead), and many project to ipsilateral optic tecta. As in ast, many axons misroute before reaching the tectum, with some crossing the midline in the PC, and others crossing rostral to the chiasm (arrow). L or R eye, left or right eye; LOT or ROT, left or right optic tectum. Neuron 2002 33, 205-217DOI: (10.1016/S0896-6273(01)00579-7)

Figure 7 slit2 and slit3 Expression in the Ventral Diencephalon Suggests Roles in Guiding Retinal Axons (A and B) Lateral (A) and ventral (B) views of slit2 mRNA expression (blue) and anti-acetylated tubulin (AT) staining (brown) at 36 hpf. Anti-AT labels all axon tracts; retinal axons are not seen as a distinct tract, but course immediately rostral to the post-optic commissure (POC). (A) slit2 is expressed in the optic stalk (OS) and along the rostral margin of the optic recess (OR). (B) slit2 is expressed in the OR and along the midline rostral to the OR, across the path of the anterior commissure (AC). (C and D) Lateral (C) and ventral (D) views of slit3 expression and anti-AT staining. slit3 is expressed in a patch of nonneural tissue overlying the optic stalks, but not in the optic stalks themselves. In the ventral forebrain, slit3 is expressed in three domains: rostral to the AC, between the AC and POC, and caudal to the POC. (E) Model for Robo-Slit regulation of retinal axon pathfinding in the ventral forebrain. mRNA for Slits is expressed in domains bounding the retinal pathway. Diffusion of Slit protein creates two opposing gradients, with the lowest concentration along the pathway. Gradients of Slit (blue) on either side prevent a normal Robo2-expressing wild-type (wt) axon (red) from leaving the pathway. A wild-type axon that happens to express a slightly lower level of Robo2 (orange) can wander up the Slit gradient, but eventually encounters a concentration to which it responds, then collapses and retracts. An ast axon (yellow), lacking functional Robo2, is unresponsive to Slit, and thus wanders uncorrected off the normal pathway. Scale bars: (A) and (C), 50 μm; (B) and (D), 25 μm. Neuron 2002 33, 205-217DOI: (10.1016/S0896-6273(01)00579-7)